AUSTIN, L.G.,MENACHO,J.M.and P,EARCY,F.

A general model for semi-autogenousand autogenous

milling. A P C O M 87. Proceedings of the Twentieth International Symposium on the Application of Computers
and Mathematics in the Mineral Industries.Volume 2:Metallurgy. Johannesburg,SAIMM, 1987. pp. 107 - 126.

L.G. AUSTIN, J.M. MENACHO and F. PEARCY

Pennsylvania State University, Philadelphia, Penn ., USA
This paper was the subject of a cross-disciplinary presentation under the chairmanship of
Dr A.M. Edwards

The paper summarizes the current state of development of simulation models

for SAG and FAG mills, and gives the results oC recent investigations of the
physical processes occurring in autogenous grinding. The breakage process
is treated as the sum of three regions of breakage actions: normal breakage

I caused by nipping of particles between media (steel balls or pebbles); abnormal

breakage caused by media when the particle or lump is too big in relation
to the media to be readily nipped; and self-breakage resulting from the chip-
ping fracture and abrasion of the tumbling action of rock lumps. Each region
of breakage action has associated specific rates of breakage and primary
progeny fragment distributions.
A simplified form of the model was used to predict the performance of
an 8 m diameter SAG mill with L/D = 0.5 grinding a copper ore, and predicted
maximum capacity and minimum kWh/ton at about 6% ball load at 25%
total filling. Two FAG mills of L/D = 2 were necessary to give the same
capacity, and the simulations indicated a lower kWh/ton for these mills. For
this ore, both systems were technically feasible.

Introduction
Although the use of simulation models for it and then use an approximate simplified
the design of ball mills has advanced form of the models to compare the
significantly in recent years,(') progress performance of a typical SAG mill design
has been slow in applying the concepts of (LID = 0.5) with that of a representative
specific rates of breakage and primary FAG mill with LID = 2. It is assumed that
breakage distributions in the construction the reader is familiar with the concepts
of simulation models (2-4) for semi- and symbblism of construction of mill
autogenous (SAG) and fully-autogenous (FAG) models. (1)
mills. This is largely because the Mill models
physical processes of breakage occurring in
The basic mass balance
these mills are more complex than those in
. ball mills. Several recent papers by It is assumed that this type of mill
Austin and co-workers (5-9) have approximates to a fully mixed reactor where
investigated certain aspects of this the grate acts like a size classifier to
problem. prevent large material leaving the mill.
In this paper, we will first summarize The simple concept of residence time
the current state of development as we see distribution loses meaning in such a system

A GENERAL MODEL FOR SAG AND FAG MILLING 107

because the residence time is 'determined by it passes through the grate (see later).
the rates of breakage of feed material to Therefore, the system is treated as shown
less than the grate size, so it is a in Figure 1, with material rejected by the
complex function of the feed size grate thrown back into the mill charge with
distribution and specific rates of an equivalent internal circulation ration
breakage. However, it is still convenient of C'. Since Ff=(l+C')F, the mass balance
to define a mean residence time by solid rrf the apparent mill feed is
hold-up W divided by solid feed rate, (l+c')~f; = Ff.
1
+ F(l+Cf)wici
T=W/F. The basic mass balance is simple to
where the value of c is the fraction of
write: i
size i material returned to the mill. The
i- 1
Fp.=Ff .+(W C b S.w.)-WSiwi, i=1,2,. . ..n value of C' is defined by
1 1
j= 1 i,j J J cl=c ~ ( l + ~ ' ) w ~ c ~ /F(~+c')w~(~-c
C .)
1
i i
or

i>1
1+C' = 1/C wi(l-ci) [21
where p is the fraction of mill product in i
i
size class i; f. is fraction of feed in Applying Eq. [l] to this system using w =
I i
size class i; S. is the specific rate of
I. _
p!1 gives
breakage of size i material; b is the
i,i- i- 1
fraction of material broken from size class f i + ~C b S.w
i,j J j
j which appears in size class i; w is the w =
i )i * T i=1,2,..n [3]
1 1
fraction of hold-up of size class j; and n
is the total number of size intervals. The equation is readily converted to the
There is evidence that material less than close-circuit form.
the grate size undergoes classification as The hold-up in the mill of sizes 1es.s

r - - - - -1
I GRATE
I I CLASSIFICATION
I
I W
F fi F'.
' fi
e

I wi
I 1
1
I
1
I
I C' II
I
OVERALL M I L L
FIGURE 1. Illustration of grate classification treatment as an exit classifier

METALLURGY: MODELLING
than the grate opening depends on the ease abrasion in addition to normal breakage
with which slurry flows from the mills: processes, and abrasion follows a different
the level must increase in order to permit set of laws than breakage. Thus, Austin
slurry to flow out at a higher rate. Thus, and Menacho (I1) have formulated the overall
the system is only precisely defined if a balance on size i as 'Rate of size i
mass transport relation is used. We will material out= (rate of size i material in)
use the empirical function + (net rate of material wearing in to size
N
fs/fso = (FV/FVo) m [41 i by abrasion of rounded pebbles through
where f is the fractional filling of the the upper and lower limits of size interval
S
mill by slurry, with slurry defined by i) - (rate of loss of fragments by abrasion
sizes less than the grate opening. F is a
v
of pebbles in size interval i) + (rate or
volume flow rate and N is an empirical production of size i material from the
m
constant. In terms of the hold-up of products of abrasion of larger sizes) -
material of sizes less than grate opening (rate of production of size i material by
in a mill of effective volume V, fracture) + (rate of production of size i
material from the products of fracture
breakage of larger sizes)'.
Considering abrasion according to a
where C is the volume fraction of solid in linear wear law it is readily shown(7,121
S
the slurry;
's
is the true density of the
solid; and i is the interval number
that the net rate of increase of mass due
to pebbles of equivalent radius r wearing
g
corresponding to the grate size. The value in and out of a differential size element
2 2
of Fv; is a standard flow rate which gives dr is W K ( ~ P/dr )dr, where P is the
a standard filling of f
SO
F
vo
.is expected cumulative mass fraction size distribution
(lo) to be related to mill size by of hold-up W and K is the linear wear rate
(LIT). Similarly, the mass loss from the
intervals is (3~/r)(dP/dr)dr. Incorpor-
where @ is fraction of critical speed and ating these with the usual first-order
C
A is the fraction of mill cross-section mass-rate balance due to fracture gives
g
which is open grate area. The simultaneous
solution of Eq. [4], [5] and [6] gives

where a is the fraction of material

ij
abraded from size j which appears in size
For given values of f .,S. and c interval i.
i' bi,J J i
there is only one value of T which will The value of K is not necessarily
satisfy both Eq. [3] and [ 7 ] . constant over a wide size range but it can
be assumed that it is approximately
A more complex mass balance
constant within a size interval; Kis thus
i
The problem with the simple mass balance defined as the mean value of K for particle
given above is that it considers overall sizes x <2r<x.. Then
i+ 1 1
first-order S values, whereas there is the
i
possibility of a substantial component of

A GENERAL MODEL FOR SAG AND FAG MILLING 109

 i
Also, by definition /i+l (dP/dr)dr=wi, so it grate classification is then given by

convenient to define a mean value ;.by the p 1.=(wi) (l+C1) (1-ci) . For closed circuit
1
the values of f. are readily replaced in
1
product of some factor u and the upper size
- terms of make-up feed g
i
using the si
of sieve size i, r,=uf and make the reason-
1 parameters for the external classifier.
i
able approximation f+l(~/r)(dP/dr)dr ' FAG mill model
-
~ . w . / r ~ .The derivative dP/dr is approxi- The FAG mill model is essentially
1 1
identical to that to the SAG mill model
mated by (d~/dr).~w /(X~-~-X~)
1 i-1
given as Eq. [8] except that it is applied
wi-l/xi-l( 1-R) , giving as an equivalent series of reactors(899)
with various degrees of recycle of the
material rejected back to the mill by the
. .
1 1
grate, as indicated in Figure 2. A grate
classification action is necessary to
retain the pebbles in the mill. Of course,
the calculation of the effective overall S
where R is the ratio of lower to upper i
values will not include a term for breakage
sieve size of the size interval (=&).
by balls. In addition, the mass transport
It is convenient to define the specific
constant k will correspond to that for a
rate of loss of material by abrasion to m
long L I D ball mill and not a short SAG
fragments as a fraction of the specific
mill.
rate of fracture; letting
i- 1
Y ~ = ( ~ K ~ / U X )~ISi,
R putting u/3(1-R)=K,
Calculation of S and B values
and setting p!=w. gives as before
1 1
Breakage by balls and pebbles
i- 1
1~ J
i+KYi-lTWi-lSi-l++ <=IC (a..y.+b. In order to allow for different ball and
1~.)S.w
J j
JPA
Wi= (l+C1)(1-c,)+ T~S,[~+Y,(~+K) 1 1 pebbles sizes in the mill, the values of
[81 specific rates of breakage of smaller sizes
to be used in place of Eq. [3] if abrasion by impact from media were calculated from
processes are significant. The size the equations developed for ball mills. (1)
distribution of the product stream from the The relations between the specific rates of
GRATE
CLASSIFICATION
r - - - - - - - - - - - - - -
1
I
I MILL I MILL 2
- -MILL m
I
I
F
fl
I
I * WI, I
TI
. , c
Wi,2
r2
----- Wi,m
Trn
- I

I
I
I "IC' e&' e,C1
C'
I
L----------------,
----
, --1
,
OVERALL MlLL

FIGURE 2. Reactors-in-seriesformulation for a F A G mill with grate classification

METALLURGY: MODELLING
breakage determined using a single ball mixture of balls can be calculated from
size in a laboratory test mill of diameter
D and those for a mixture of balls in a
Bi,j-
-
:nkSj ,kBi,j ,kl: ?cSj ,k [I21

T Figure 4 illustrates the decrease of

larger mill of diameter D are
breakage rates as the particles become too
large to be nipped properly by the balls.
In Region 2, the breakage of the material
in a batch rate test becomes non-first
order (abnormal) because it consists of a
mixture of normal fracture caused by a
with
direct impact and chipping caused by a
glancing, less violent impact. Figure 3
shows the variation of overall B values as
particle size increases with respect to the

2.3 ball size. It is clear that chipping

D N1 3.81 A 1+6.6J becomes more important as the size
C =(-I (-1 (- T- ) exp [-c(U-UT) 1
D~ 1+6.6J2s3 increases.
$ -0.1 l+exp [15.7($cT-0.94)]
(-+cT-O.l ) ( l+exp [15.7($c-0.94)] )
[loci Equations [9] and [lo] were used for both
balls and pebbles, with the values of the
characteristic parameters and pT
where A=O for SAG milling; A=O for D<3.81 m
adjusted by
and = 0.2 for D>3.81 m for ball milling or
density pebbles pp
long LID FAG milling. pebbles = balls x
T T density balls p
The subscript T denotes conditions in the
8131
test mill, and m is the mass fraction of
k uT pebbles = pT x pp/pB [ 141
the kth size class of balls in the
full-scale mill: No, N1, N2, N
are The value of J to be used in Eq. [lOc] is
3
empirical exponents based on experience based on the total media filling which
with ball mills. The value of U is the defines how media tumbles in the mill, that
fractional interstitial filling of the is, it must be the sum of filling by balls
media voids by particles and J is the and pebbles, J=J +J Calculation of
B P'
fractional filling of the.mill by media. breakage due to balls using this J would
The values used for the exponents were only be valid if all the media were steel,
N =0.5, N =0.2, N =1.0 and N =1.0, and the so the value is corrected by the factor of
1 2 3 0
cushioning factor c was taken as 1.3. fractional volume J ~ / J ,s (B) i=(~B/J) C
mkS(dk)i. Similarly, S(P)i = (Jp/J) C
The cumulative primary progeny fragment
w.S (x.) where w is the mass fraction of
distributions (B .) for normal breakage by J ~i j
~ Y J
pebbles of size interval j in the tumbling
a single size of ball of class k have been
charge. It is assumed that all rock larger
found to be fitted by the empirical
than the grate size, plus balls,
function
contributes to the media filling level of
the mill.
The term exp [-c(U-UT) ] in Eq. [lot]
as illustrated in Figure 3. For comparison represents the decrease of specific
purposes, the overall B value for a breakage rates when the media void spaces
i,j
A GENERAL MODEL FOR SAG AND FAG MILLING 111
 RELATIVE SI Z E , XI / ~1
FIGURE 3. Typical variation of the breakage distribution values for large particles. Dry batch grinding
of quartz in a 0.6 m diameter mill. U = 0.5, J, = 0.2, 26.4 mm diameter balls, 75% of the
critical speed

are overfilled. This e f f e c t s t a t e s t h a t Self-breakage

e x c e s s i v e powder i n t h e media c u s h i o n s t h e A s p a r t i c l e s i z e i s increased the
breakage a c t i o n on t h e powder. I t seems p a r t i c l e s a r e n o t nipped by tumbling b a l l s .
reasonable t h a t the cushioning a c t i o n However, they e v e n t u a l l y become b i g enough
depends on t h e media s i z e , t h a t i s , a 10 mm t o b r e a k by t h e impact of t h e i r own f a l l ,
d i a m e t e r lump w i l l b e media t o s m a l l s i z e s i n t h e s t r e a m of tumbling r o c k and b a l l s .
b u t should be counted i n t h e c u s h i o n i n g Then t h e breakage r a t e s i n c r e a s e w i t h
powder f o r p e b b l e s of 100 mm d i a m e t e r . i n c r e a s e d lump s i z e due t o t h e i n c r e a s e d
Using t h e Weymont ( I 3 ) voidage (n) factors a impact f o r c e , g i v i n g Region 3 of F i g u r e 4 .
simple method of d e f i n i n g powder i s t h a t The t r a n s i t i o n from Region 2 t o Region 3 i s
a l l m a t e r i a l of s i z e l e s s t h a n 0.125 of t h e o b t a i n e d by simple a d d i t i o n ,
media s i z e i s c o n s i d e r e d t o be powder. Si=S (B) i+S ( P ) i+S (S) where S (B) and S (P)
i
T h i s c o r r e s p o n d s t o a voidage f a c t o r of a r e breakage due t o n i p p i n g by b a l l s and
s i z e i w i t h r e s p e c t t o media s i z e k of 0.5. p e b b l e s (which become s m a l l a t l a r g e
Then t h e term i n E q . [lOc] becomes s i z e s ) , and S(S)i h a s been termed s e l f -
exp[-c(U -U ) ] , s i n c e t h e e f f e c t i v e v a l u e breakage. The v a l u e s of S ( S ) . a p p e a r a l s o
k T 1
of U depends on t h e media s i z e b e i n g t o f o l l o w a power f u n c t i o n
considered.

112 METALLURGY: MODELLING

 - 1.0 .
I* REGION I REGION 3 /
0
C
3 SI = an( xl /xolde
C

S I E V E SIZE, mm

FIGURE 4. Typical shape for the sum specific rate of breakage in SAG

a
S
S(SIi = s(~i/~o) [I51 gives
CL
S (S) ST(~i/~o)S ~ S l ~ S 2 [ 181
where S is smaller than the corresponding
for media, so that S(S)i is negligible N1
for small particle sizes.
where
C =
(DIDT) -
D<3.81 m
S1
The empirical equation used to allow for -
(3.81/~~)~1(~/3.81)~1-*D>3.81 m
the effect of media filling on JBP B+J PPP
C =( 1+(~~/0.4)
self-breakage is 52 1+(J /0.4)~)(
J~ P~
> exp [ -cs (Uk-UT> 1
1
S(Si a [ 161
1+(~/0.4) where c is the cushioning factor for self-
S
breakage.
Since breakage is proportional to SJ, this
gives a maximum rate of breakage at J=0.4. Experimental data
When a fraction of the media is steel balls Ball x mill breakage
it seems logical that the rock in the The material used for the simulations was
tumbling stream will receive a a composite sample of copper ore from the
proportionate fraction of heavier impacts. Los Bronces mine in Chile. The standard
Therefore, the breakage rates were method of determination of breakage
corrected by parameters in a small laboratory mill as
S (S) i=(JB~B+Jp~p)
/JppP [171 described by Austin, Klimpel and Luckie (1)
Thus, although the breakage rate of a given was used, giving the characteristic
size of tumbling rock is termed parameters shown in Table 1.
self-breakage it is accepted that the size
and density of the rest of the charge has Self-breakage parameters
an influence on this breakage. The relations governing the variation of
It is expected that the specific rates of self-breakage parameters with mill
self-breakage will scale with mill diameter conditions are not as well developed as
in essentially the same way as breakage by those for ball milling, so a series of
balls as in Eq. [lo]. Combining equations tests was performed using a readily

A GENERAL MODEL FOR SAG AND FAG MILLING 113

TABLE 1 . Breakage parameters for copper mm/min. Sudden changes in weight
ore determined in a laboratory represented chipping of fragments from a
mill lump, whereas complete disappearance from
Conditions Parameters the size interval represented a
D = 194 mrn a = 0.95 disintegrative fracture.
d=27mm A = 3.3 The results indicated that pure abrasion
J = 0.20 = 1.65 mm gave relatively low rates of weight loss,
B
U = 0.5
-1 so abrasion and chipping were combined to
= 1.0 min
T
Volume solid X = 40 y = 0.70 give an overall chipping-abrasion rate.
3 Figure 6 illustrates the process in terms
Ore density = 2.77t/m B = 4
Mill speed = 75X C.S. @ = 0.36 of the mass balances involved in the

- - - -
derivation of Eq. [ B ] . Figure 7 shows a
These values were also used to predict typical result of the balance between
normal breakage by pebbles using the average mass loss by fracture by fragments
conversions of Eq. [ 1 3 ] and [ 1 4 ] . from chipping-abrasion and by cores wearing
into the next smaller size internal. It is
availaule and homogeneous material, a white concluded that the proportion between these
crystalline quartz from the mine of the mechanisms remains constant during the
Castastone Company, North Carolina, with a process. Table 2 shows the variation of
Bond work index of 19 kWh/ton metric. these ratios for a series of tests. The
These tests were with particular reference ratio of the mass lost as chipped fragments
to elucidating the action of fracture to that lost by cores for the 4J2 size
breakage, chipping and abrasion during intervals is expected to be about 33% since
autogenous grinding. The test mill used the distribution of mass is linear over the
was 0.6 m in diameter and 0.3 m long, 442 interval. (7)
fitted with 20 lifters. An adsorbable dye However, Figure 8 shows that the overall
was used to number the feed material, and specific rate of breakage of traced lumps
the weight of individual lumps followed as decreases with time even though the
a function of grinding time in batch conditions in the mill are kept
grinding tests. The weights were expressed approximately constant. The equivalent
as equivalent spherical radii to enable the decrease can also be seen in the pure
calculation of abrasion rates. abrasion rates in Figure 5 . The physical
Figure 5 shows typical results. To avoid process involved is quite clear. The raw
the variation of parameters caused by the feed lumps are irregular and some have a
cushioning action of fine material and the relatively high surface area per unit
decrease of mill charge, the tests were weight, thus they abrade and chip
performed dry, the fine material was relatively rapidly. Chipping-abrasion
removed after short grinding periods, and causes the lumps to become rounded into
the equivalent weight of unmarked fresh pebbles, which chip-abrade more .slowly.
lumps added to preserve the desired J The direct proportionality of Figure 7
value. This type of test enables the shows that the fracture component follows
linear decrease of weight due to abrasion suit, so that the system behaves like a
to be readily determined, and values were mixture of weaker material which disappears
averaged to obtain mean abrasion rates K more rapidly, leaving stronger rounded

METALLURGY: MODELLING
 GRINDING T I M E , min

FIGURE 5. Change of equivalent spherical radius with time for 86 individually-traced lumps within the
sieve range 63 x 53 mm; mean R, = 31 mm, upper r, = 35 mm, lower r, = 27 mm (D =
0.6 m, fine material removed after grinding period)
 CHIPPING -ABRASION FRACTURE

i-2

i-l

FIGURE 6. Illustration of mass balances of chipping-abrasion and fracture in autogenous breakage

0.6 p e b b l e s , which d i s a p p e a r more s l o w l y .

F i g u r e 8 a l s o shows t h e s t r o n g i n f l u e n c e
of t h e a c c u m u l a t i o n o f f i n e m a t e r i a l , s i n c e
t h e b r e a k a g e r a t e s w i t h U=0.8 a r e f a r
s l o w e r t h a n f o r U=0.3. F i g u r e 9 shows a
t y p i c a l r e s u l t considered a s a simple
binary mixture: t h i s is surely an
approximation b u t i t i s s u f f i c i e n t f o r our
p r e s e n t purpose. The s q u a r e s a r e showing
t h a t t h e breakage r a t e of f r e s h rock i s
proportional t o the f r a c t i o n of f r e s h rock
p r e s e n t , t h a t is, i t s breakage follows a
first-order law. F i g u r e 10 shows b r e a k a g e

0 r e s u l t s i n t h e presence of a v a r y i n g
0 0.5 1.0 q u a n t i t y o f b a l l s t o i n c r e a s e t h e mean
TOTAL FRACTIONAL WEIGHT LOSS d e n s i t y of t h e load. As e x p e c t e d , t h e n e t
m i l l power i s d i r e c t l y p r o p o r t i o n a l t o t h e
FIGURE 7. Contribution of individual breakage mechanisms
to the total fractional weight loss. Fully mean d e n s i t y o f t h e l o a d , and t h e s p e c i f i c
autogenous test in a 0,6 m diameter mill. Quartz
breakage r a t e s a r e a l s o p r o p o r t i o n a l ,
lumps of 63 x 53 mm in diameter, U = 0, J =
0.30, 4, = 0.70 v a l i d a t i n g Eq. [ 7 ] .

METALLURGY: MODELLING
 GRINDING .TIME, min
FIGURE 8. Rates of self-breakagefor 26.5 x 22.4 mm copper ore. Dry batch autogenous grinding D
= 0.6 m, 4, = 0.7, J = 0.2,powder minus 100 mesh

TABLE 2. Deconvolution of chipping and fracture (quartz; D=0.6 m)

Steel balls Pebbles Powder Relative Breakage

Test Size Size Size U Fracture Cores Chips Chips/

mm mm mesh (Cores+Chips)
J~ J~

'

Test 19 is a mixture of 6 different sizes of JP = 0.30 total.

Figure 11 shows that the primary breakage from those of normal fracture, which is to
distributions are approximately normalized be expected since the component of fracture
with respect to the breaking size, but are is relatively small (see Table 2). The
very different for the fresh feed and the shape clearly corresponds to a mixture of
rounded pebbles. In addition, the shapes fracture plus the cores and fragments
of the distributions are quite different produced by chipping-abrasion. An estimate

A GENERAL MODEL FOR SAG AND FAG MILLING 117

 -
-
-
-

- \ S, FAST=0.42
min-I -
0 EXPERIMENTAL
- SIMULATED

GRINDING T I M E , min
FIGURE 9. Determinationof combined rate of breakage-chippingfor 63 x 53 m m fresh feed (D= 0.6 m,
J = 0.2, 4, = 70% of critical speed,quartz, no accumulation of fines)

of the size distribution of abrasion-

chipping fragments was made by rotating the
mill at a low fraction of critical speed
(45%) to give a flat angle and, hence, to
reduce the impact forces of the tumbling.
This gave the result shown in Figure 12 as
self-abrasion.
It was also found that in a mixture of
lump sizes, the presence of larger lumps
increased the breakage rates of smaller
lumps while the presence of smaller lumps
decreased the breakage rates of larger
lumps. No quantitative relation for this
effect has yet been deduced.
The non first-order nature of the self-
breakage as indicated in Figures 8 and 9
was found in all tests and complicates the
analysis. The equivalent results for the
copper ore were expressed as mean specific
0.01 I 1 I 1 I 1 1 1 1 1

1 2 3 4 5 6 78910 rates of breakage (I4) defined by

MEAN DENSITY OF THE MILL LOAD,
metric ton l m 2
FIGURE 10. Variation of mill power and specific rates of self-
breakage with density of the mill road:
On this basis, the value of c was taken to
53 x 45 mm quartz in 0.6 m diameter mill; S
45 mm steel balls; J = 0.30 be 1.3 and a =l. The value of ST was
S

METALLURGY: MODELLING
 -
- I I I 1 1 1 1 1 1 ) I I 1 1 1 1 1'1 I
-
-
-
-
-
-
AGED

-
-
-
-
-
-
FEED SIZE, mm -
45 x 38 -
A 5 3 x 45 -
&4 QUARTZ
o 6 3 x 53
- -
-
-
--
- -
I I II I I 111 I11111 I I 1 1 1 1 1 1

R E L A T I V E PARTICLE S I Z E , xi / x j
FIGURE 11. B values from fully-autogenous batch grinding tests in 0.6 m mills; J = 0.30, U = 0. 4,
= 0.7

RELATIVE SIZE , xi/xj

FIGURE 12. Cumulative primary breakage distribution values for 53 x 45 mm copper ore. Dry batch
autogenous grinding in a 0.6 m diameter mill

A GENERAL MODEL FOR SAG AND FAG MILLING

 determined by back-calculation(8' from filling. Thus, the SAG mill can pass large
steady-state continuous pilot-scale data to quantities of slurry without overfilling
be 0.7 x min
-1 for a test mill with slurry, by comparison with an overflow
diameter of 1.8 m and ~ / ~ = 0 . 3at
, JT=0.2, ball mill.
@ =0.77 and approximately 40 volume 2 of The comparison of size distributions
C
solid in slurry leaving the mill. within the pilot-scale to those leaving the
mill gave a grate classification function
Mass transport relations and grate classification of the form
X
The continuous pilot-scale tests were .
ci = ~ / [ ~ + ( X ~ ~ Ig1X ~ ) [201
also used to estimate the value of Fvo in
Eq. [6]. This gave a value of A k at For the 12 mm (half inch) grate opening,
g m
£so
-1 0.5
=0.25 of 0.37 min m .
On the other the values of the characteristic parameters
hand, the mass transport relation given for l l and X =1.3. However, it
were ~ ~ ~ = l . mm
! g
overflow ball mills(10) gives k A =0.5 h-I must be understood that this action may be
n
u.3
c m g
m for a solid filling level different in a full-scale mill where the
corresponding to approximately U=l at grates are kept freely open by an adequate
J=0.38, which is about the same slurry discharge mechanism. In this case, the

TABLE 3. Values used in simulations

B values for various breaking sizes

By pebbles and balls
Feed Self-Breakage Classifier
Size Size % Minus Pebbles Balls selectivity
Interval pm size 1-3 4 5-26 1-3 4-11 12-26 12-26 s
i

120 METALLURGY: MODELLING

action of the grates can be considered as of 50% of 76 mm (3 inch) and 50% of 100 mm
close to ideal classification, which can be (4 inch) diameter balls. The mill was
closely approximated by x50=grate size and closed with the external classifier
X =4. selectivity values given in Table 3.
g
The variation of mill capacity with
Full scale simulation hold-up and ball charge is shown in Table
SAG mill: L/D = 0.5 4, and Figure 13. The charge load of 0.25
The mill simulated was a nominal 28 feet volume fraction was taken as optimum since
by 14 feet mill, giving D=8.2 m, ~/D=0.5 it gave a region where the capacity is not
and an effective volume of'230 m
3
The . sensitive to ball load in the region J =4%
B
feed and overflow trunnion diameters to 84. The power equation used was that
represented a filling level of 29% to the given by Austin (16)
level of overflow, and the design criteria 0.
m = K(D~'~)(LID) (J) (1-1.033) ($c-2mc)~c
was 25% that is, J=0.25, and 76% of P
critical speed. The cumulative primary ,kw [211
breakage values estimated from laboratory where p is the overall density of the
C
batch tests are given in Table 3, and were charge, including balls and slurry. This
entered as matrices into the program. The was calculated using the bed porosity
simulator was supplied with the feed size obtained from the use of the Weymont
distribution also shown in Table 3, and voidage factors, as described by Austin.
simulations were performed for nominal The value of K was taken as 10.6 for
fractional hold-up(JT) in the mill covering dimensions in meters and p in metric
C
3
the range 0.2 to 0.4, using a ball charge tonslm , to produce the mill power quoted

TABLE 4. Results of SAG mill simulations

Hold-up Product size

Circulation Capactiy W distribution X <
J~
J Ratio Q, tph tons 35 mesh 400 mesh kWh/t

A GENERAL MODEL FOR SAG AND FAG MILLING

 of m i l l charge i n t h e s i z e r e g i o n of 20 mm
t o 60 mm i s t h u s reduced.
FAG mill: L/D = 2.0

I t was decided t o compare t h e r e s u l t w i t h

two e q u i v a l e n t FAG m i l l s each p u l l i n g h a l f
of t h e SAG m i l l power, a s p r e d i c t e d by t h e
Bond e q u a t i o n f o r b a l l m i l l s ,

where p i s normally t h e b a l l d e n s i t y b u t
b
i n t h i s c a s e was t h e o r e d e n s i t y of 2.77
tons/m
3
.LID was assumed t o b e 2.0,
@ =0.85, and t h e v a l u e of J was t a k e n a s
C
PERCENT MILL FILLING, J\
0.3. With t h e s e assumptions, t h e m i l l
FIGURE 13. Variation of predicted mill capacity with filling
and ball load for nominal 28 feet diameter by d i a m e t e r i s D = 5.6 m t o g i v e a m i l l s h a f t
14 feet long SAG mill power of 2000 kW.
by Tanaka and Tanaka. (15) The s i m u l a t i o n was performed w i t h t h e
The v a l u e s of s p e c i f i c g r i n d i n g energy i n
same breakage p a r a m e t e r s used f o r t h e SAG
kWh/ton a r e a l s o g i v e n i n Table 4 , and
m i l l i n g , t h e same g r a t e openings t o r e t a i n
F i g u r e 14 shows t h a t t h e minimum o c c u r s a t
l a r g e m a t e r i a l and t h e same e x t e r n a l
about 6% b a l l l o a d . F i g u r e 15 shows t h e
c l a s s i f i e r function. The f e e d t o t h e 5.6 m
s i z e d i s t r i b u t i o n s p r e d i c t e d by t h e model
i n t e r n a l d i a m e t e r FAG m i l l was t a k e n a s t h e
f o r o p e r a t i o n a t 8% b a l l l o a d and f o r
same a s t h a t t o t h e SAG m i l l .
fully-autogenous operation. Figure 16
No i n f o r m a t i o n was a v a i l a b l e f o r t h e
shows t h a t t h e e f f e c t of t h e b a l l s i s t o
expected mass t r a n s p o r t r e l a t i o n , but t h e
i n c r e a s e t h e s p e c i f i c r a t e s of breakage of
comparison of t h e v a l u e s f o r t h e SAG m i l l
t h e l a r g e r s i z e s , w h i l e F i g u r e 15 shows
w i t h t h o s e f o r overflow m i l l s s u g g e s t s t h a t
t h a t . t h e consequence i s t h a t t h e f r a c t i o n
t h e long L/D m i l l s f i l l up w i t h s l u r r y t o a
g r e a t e r e x t e n t t h a n t h e SAG m i l l ,
presumably due t o t h e r e s i s t a n c e t o mass
flow through t h e long l e n g t h of tumbling
charge. Thus, t h e v a l u e of f i n Eq. [4]
so
was i n c r e a s e d a r b i t r a r i l y by a f a c t o r of 3.
Table 5 g i v e s t h e r e s u l t s . The s p e c i f i c
g r i n d i n g energy i s s u b s t a n t i a l l y l e s s , b u t
t h i s r e s u l t i s s e n s i t i v e t o t h e mass
transport f a c t o r so i t i s not possible t o
a s s e r t t h i s w i t h a h i g h d e g r e e of
0 4 8 12 16 confidence. I n a d d i t i o n , i t may n o t be
PERCENT BALL LOAD v a l i d t o u s e t h e Bond power e q u a t i o n f o r a
FIGURE 14. Optimum of capacity and specific grinding
pebble m i l l , s i n c e i t p r e d i c t s a much lower
energy with ball load at total filling of 25% mill
volume (see Fig. 13) m i l l power t h a n e q u a t i o n 21.

METALLURGY: MODELLING
 S l Z E I N T E R V A L i AND SlZE

FIGURE 15. Simulated size distributions for SAG mill (see Fig. 13) with 8% ball load and 0% ball load

Discussion of results
TABLE 5. R e s u l t s of s i m u l a t i o n s of 5.6 m
d i a m e t e r FAG m i l l (LID = 2.0) It i s c l e a r t h a t t h e o r e s t u d i e d was a
c o h e r e n t o r e which would be s u i t a b l e f a r
Nominal f r a c t i o n a l hold-up FAG g r i n d i n g s i n c e i t l e a d s t o a m i l l
X minus 35 mesh i n c i r c u i t charge w i t h s u f f i c i e n t l a r g e rounded
Z minus 400 mesh i n c i r c u i t
product p e b b l e s t o be e q u i v a l e n t t o a b a l l m i l l
Output Q, t p h ( w i t h lower d e n s i t y of media, of c o u r s e ) .
Circulation r a t i o C
M i l l hold-up, t Although t h e s p e c i f i c r a t e s of breakage of
M i l l power, kW t h e s e l a r g e lumps a r e r e l a t i v e l y low,
S p e c i f i c g r i n d i n g energy,
kwh/ t r e q u i r i n g l a r g e r m i l l volume a s compared t o
b a l l m i l l i n g , t h e chipping-abrasion p r o c e s s

A GENERAL MODEL FOR SAG AND FAG MILLING 123

 I
FULLY
AUTOGENOUS \ \

SELF BREAKAGE 1

S I Z E INTERVAL i AND S I Z E
FIGURE 16. Simulated specific rates of breakage for conditions of Figure 15

produces a high proportion of fines which total fractional mill charge of 25 volume X
compensates for the low breakage rates to gives a substantial increase in capacity
give specific grinding energies and circuit (from 305 tph to 400 tph) over FAG milling
products comparable to more conventional in the large mill, with a significantly
crushing-grinding circuits. However, the finer circuit product (approximately 32%
use of 6 volume % of ball charge and a minus 400 mesh versus 33%) and somewhat

124 METALLURGY: MODELLING

lower circulating load. References
The long LID FAG mills gave similar AUSTIN, L.G., KLIMPEL, R.R. and LUCKIE,
circuit product size distributions, but P.T. Process Engineering of Size
required two mills of 5.6 m diameter to Reduction: Ball Milling. New York,
perform the same duty as the one 8.2 m N.Y., Society of Mining Engineers of
diameter SAG mill. Clearly, for this type A.I.M.E., 1984, pp. 1-561.
of ore, the choice of system depends on the STANLEY, G.G. Mechanisms in the
comparative economic calculations of autogenous mill and their mathematical
capital and interest cost and steel representation. J. So. Afr. Inst. Min.
consumption. Metal., Vol. 75, 4, Nov. 1974, pp.
There remains substantial work to be done 77-98.
to improve the model to a level of accuracy GAULT, G.A. Modelling and control of
sufficient for utility. The use of average autogenous grinding circuits. Ph.D.
specific rates of self-breakage is not as Thesis, University of Queensland,
satisfactory as the use of equations Australia, 1974.
representing a fast-slow breakage process. VANDERBEEK, J.L., HERBST, J.A.,
The current method of entry of B values is PAJAMANI, K. and HALES, L. Online
not satisfactory. The lack of precise estimation of particle size
relations to describe the variation of S distribution and volumetric filling in
and B values with mill conditions prevents a SAG mill with a view to automatic
the development of simple laboratory tests control. Presentation, Society of
to predict the values from a small amount Mining Engineers of A.I.M.E. meeting in
of laboratory work. The results are Alburquerque , N .M. , Oct . 1985.
sensitive to the level of slurry in the AUSTIN, L.G., WEYMONT, N.P., PRISBEY,
mill, but the calculation of this level by K.A. and HOOVER, M. Preliminary
mass transport relations is based on results on the modeling of autogenous
entirely inadequate information. The model grinding. Proc. 14th Int. APCOM
at the moment does not include the effect Conference. Ramani, R.V., The
of ball diameter on self-breakage. Pennsylvania State University, P.A.,
It will be interesting to program the 1976, pp. 207-226.
model for removal of material from the mill AUSTIN, L.G., WEYMONT, N.P., BARAHONA,
contents by pebble ports, followed by C.A. and SURAYANARAYAN, K. An improved
crushing of these sizes, in the size range simulation model for semi-autogenous
of 38 mm (1.5 inches) to 75 mm (3 inches). grinding. Powder Technology, Vol. 47,
At the moment, it is not possible to 3, 1986, pp. 265-283.
demonstrate the effect of using a less AUSTIN, L.G., BARAHONA, C.A. and
coherent ore because the associated MENACHO, J.M. Fast and slow chipping
variation of the self-breakage B values is fracture and abrasion in autogenous
not known: the rapid breakage of a weak grinding. Powder Technology, Vol. 46,
ore is expected to give B values with a 1, 1986, pp. 81-87.
smaller fraction of fines. AUSTIN, L.G., BARAHONA, C.A. and
Acknowledgements MENACHO, J.M. Investigations of
The work was performed under NSF Grant autogenous and semi-autogenous grinding
CPE 8303154. in tumbling mills. Proc. 1st. World

A GENERAL MODEL FOR SAG AND FAG MILLING

 Congress Particle Technology. systems, Ph.D. Thesis, The Pennsylvania
Leschonski, K., Nuremberg, Federal State University, P.A., 1979.
Republic of Germany, April 1986. 14. AUSTIN, L.G., TRIMARCHI, T.J. and
9. HOYER, D.I. and AUSTIN, L.G. A WEYMONT, N.P. An Analysis of some
simulation model for autogenous pebble cases of non-first order breakage
mills, Society of Mining Engineers of rates. Powder Technology, Vol. 17,1,
A.I.M.E. meeting in Alburquerque, N.M., 1977, pp. 109-113.
Oct. 1985. 15. TANAKA, T. and TANAKA, K. Design
10. ROGERS, R.S.C. and AUSTIN, L.G. features of a semi-autogenous grinding
Residence time distributions in ball mill and a comparison of test mill data
mills. Particulate Science and with actual operation data. Santiago,
1

Technology, Vol. 2,2, 1984. pp. Chile, Centro de Investigaci n Mineva y

193-204. Metalurgica, Primer Taller Nacional de
%
11. AUSTIN, L.G. and CONCHA, F.A. Diseno Molienda Autogena de Minerales.
simulacibn de circuitos de molienda. Preprint No. 6, Sept. 1983, pp. 1-26.
To be published. 16. AUSTIN, L.G. Chapter 14,
12. MENACHO, J.M. Some solutions for the Semi-autogenous (SAG) and fully
kinetics of combined fracture and autogenous grinding. Short course
abrasion breakage, Powder Technology, notes: Process Engineering of Size
Vol. 49,1, 1986, pp. 87-95. Reduction in Tumbling Mills, The
13. WEYMONT, N.P. The analysis and Pennsylvania State University, P.A.,
simulation of autogenous grinding September 1987, pp. 14.1-14.55.

METALLURGY: MODELLING